Abstract
Purpose.:
We investigated abnormalities of the photoreceptor inner/outer segment (IS/OS) junction layer viewed “en face” and their functional correlates in type 2 idiopathic macular telangiectasia (type 2 MacTel).
Methods.:
Segmentation and “en face” imaging of the IS/OS lines in spectral domain optical coherence tomographic (SD-OCT) volumes were performed manually. Mesopic retinal sensitivity thresholds were determined using a Nidek MP1 microperimeter. “En face” SD-OCT images and microperimetric data were superimposed over images of the fundus. Retinal structure and characteristics of type 2 MacTel were analyzed, and associations of structural changes with function were investigated.
Results.:
We examined 49 eyes of 28 patients (mean age 62.6 ± 9.4 years). Total IS/OS break area ranged from 0.04 to 2.23 mm2 (mean 0.63 mm2, SD 0.53 mm2) and 0.03 to 1.49 mm2 (mean 0.49 mm2, SD 0.42 mm2) in right and left eyes, respectively. A correlation between fellow eyes was present (Spearman correlation ρ = 0.770, P < 0.01). An assessment of the repeatability of IS/OS lesion area measurements (n = 19 eyes) revealed an intra-class correlation coefficient of 0.99 (95% confidence interval [CI] of 0.975–0.996). Retinal areas corresponding to an IS/OS break showed a mean retinal sensitivity of 8.3 ± 5.8 and 8.7 ± 5.7 decibels (dB) in right and left eyes, respectively. Mean sensitivity over retinal areas outside the lesion was significantly higher, 17.0 ± 3.3 and 16.7 ± 3.6 dB in right and left eyes, respectively (paired t-test, P < 0.01). Mean aggregate retinal sensitivity loss was 33.5 ± 30.4 dB (n = 40), correlating well with IS/OS lesion area (Pearson correlation coefficient = 0.848, P < 0.01).
Conclusions.:
“En face” OCT imaging of the IS/OS junction layer provides a functionally relevant method for assessing disease severity in type 2 MacTel.
A total of 74 OCT volume scans was processed for this study. The Q factor (Topcon-specific quality factor reflecting signal strength) ranged from 25.0 to 75.3 (mean 56.9, SD 11.5). To assess the utility of “en face” imaging in real-life situations, no scan was excluded based on a low Q-factor alone. Motion artifacts due to microsaccades and/or drifts larger than 100 μm were present in 14 volume scans. In 4 cases, these were outside the region of interest (ROI) and were ignored. Eye movements were parallel to B-scan direction within the ROI in 10 cases and misalignments between B-scans could be corrected. In 6 cases the eye movements had a major component perpendicular to the B-scan. In 2 cases this resulted in a resampling of the same retinal area two or more times; in one case this could be corrected. In 4 cases vertical saccades resulted in scans of disparate retinal areas, and these scans were not used. One volume scan intended for assessment of variability was discarded due to axial eye/head movements during the scan. Uneven field illumination was noted in 5 scans.
The IS/OS break appears as a darker area within the en face image against the background of the highly reflective IS/OS layer (
Fig. 1). The edges of the break varied from sharp to indistinct, corresponding to a sudden discontinuity or gradual loss of reflectivity respectively. Near-black (low-reflective) areas were apparent internal to the edges of the lesion in some cases. These corresponded to cross-sections of outer retinal empty spaces at the level of the IS/OS junctions. Structures with a reflectivity similar to that of the IS/OS also were apparent within some lesions. In some cases these corresponded to islands of preserved IS/OS, but more frequently to the cross section of an area with pathologic vertical restructuring of the retina. Retinal layers between the outer plexiform layer and the RPE in these areas seemed to be absent, while the disorganized outer plexiform layer and layers interior to it gave the impression of “collapsing” onto the RPE (
Figs. 1C–
E).
Total IS/OS break areas ranged from 0.039 to 2.229 mm
2 (mean 0.627 mm
2, SD 0.529 mm
2) in right eyes and from 0.027 to 1.494 mm
2 (mean 0.492 mm
2, SD 0.422 mm
2) in left eyes. A strong positive correlation between lesion area in left and right eyes was found (Spearman correlation
ρ = 0.770,
P < 0.01,
n = 42,
Fig. 2A). When repeatability of IS/OS lesion area measurements was assessed by processing two consecutive scans of 19 eyes the intra-class correlation coefficient was 0.99 (95% confidence interval [CI] of 0.975–0.996).
Topographically, the IS/OS lesion was located typically on the temporal side of the fovea, in 65% reaching the center of the foveal depression. Mean radial distance of the nearest lesion edge from the foveal center was 158 μm (SD 220 μm, range 10–890 μm) in cases sparing the foveal center, and 126 μm (SD 125 μm, range 15–560 μm) in cases involving the foveal center. A positive correlation was found between break area and degree of involvement of the foveal center in right eyes (Spearman correlation ρ = 0.44, P < 0.05, n = 25) as well as in left eyes (ρ = 0.58, P < 0.01, n = 24).
An oval ring or ring segments with a backscatter lower than that of the IS/OS was apparent around the IS/OS break in 25 eyes (51%,
Figs. 2B,
2C). In 4 cases this ring was accompanied by one of higher reflectivity than the surrounding IS/OS layer (
Fig. 4).
BCVA.
Microperimetry.
Fixation Stability.
In our study, we isolated backscatter information from the level of the line attributed to the photoreceptor IS/OS junctions to develop a more sensitive structural predictor of vision loss in type 2 MacTel. By 3D processing of standard volume scan data from a commercially available SD-OCT device and imaging the data “en face,” we demonstrated the characteristics of a break in this layer. This break was closely associated with a loss of mesopic retinal sensitivity. These findings are significant not just to allow more sensitive monitoring of the condition, but also as an outcome in trials of potential treatments.
“En face” OCT imaging of the retina is intuitive for the examiner, and permits 2D assessment of lesion extent, topographical analysis, and close comparisons with other imaging modalities. Prototype OCT systems using a C-mode configuration, performing single planar, 2D transverse scans
34,35 (i.e., “en face,” in the coronal plane) as well as 3D volume scans
36 have been developed. It also has been demonstrated in experimental OCT systems producing 3D volumes of traditional axial scans that OCT “en face” fundus images can be created by summing reflectivity data in A scans,
37 and that backscatter intensity information specific to selected slabs or individual retinal layers also can be extracted.
19,38
In our study we used volume scans from a standard, commercially available SD-OCT machine that was not equipped with a real-time eye-tracking system. Since fixation stability in type 2 MacTel patients typically is affected early, motion artifacts may be a major source of error. We acknowledge that in the absence of vascular landmarks, in some “en face” images, near the foveal center, it may not be possible to detect artifacts due to minor eye movements. However, in 19 of 20 double scans performed consecutively in our study, lesion area measurements showed a high degree of reproducibility between scans and over time. In one case, one scan of a pair had to be discarded due to obvious inappropriate axial movement by the patient. Performing two consecutive scans routinely may reduce the risk of undetected motion artifacts further. Machines with active eye-tracking do, however, have a significant advantage.
We performed manual segmentation of the IS/OS line. Several methods for automated segmentation of retinal layers have been published previously.
27,39–42 However, extensive discontinuities in and reduced backscatter from the IS/OS are common near the foveal center in type 2 MacTel, which may cause errors in automated boundary detection. Manual segmentation was deemed to provide more accurate results.
Type 2 MacTel is associated with evidence of retinal neurodegeneration that is integral to the disorder. Imaged “en face” at the level of the IS/OS line, the most prominent feature of eyes with type 2 MacTel is the break in the IS/OS line, which appears dark against the highly reflective background of the IS/OS. Its boundaries may be distinct but also indistinct. Where the edge is indistinct, the IS/OS line often appears in B-scans thinner and deviates toward the choroid from its expected location, which may correspond to a shortening of the photoreceptor outer segments and a disorganization of the IS/OS junctions. An attenuated signal from the IS/OS also may in itself be a sign of photoreceptor damage.
43 It is possible that the “IS/OS line” may not be representative of the IS/OS junctions. Recent studies indicate that this line may, in fact, align with the ellipsoids of the photoreceptor inner segments.
43–45 Its functional relevance, however, is well established.
Within the area of the break, regions with very low backscatter may be present. These correspond to the cross sections of outer retinal spaces with low reflectivity. In B-scans, these spaces give the impression of degenerative cavities, typically without signs of internal pressure. The material within these spaces does not take up fluorescein even in the late phase of fluorescein angiography.
Insular areas with a reflectivity closer to that of the IS/OS also were apparent within breaks. These may correspond to areas with preserved IS/OS, but also to the cross sections of an abnormal retinal tissue with a vertical reorganization, where retinal layers between the outer plexiform layer and the RPE seem to be absent and the structurally disorganized remaining layers give the impression of “collapsing” onto the RPE. In cases where the break did not reach the foveal center, this structure always was seen on its temporal side. Based on “en face” images alone it may not be possible in all cases to determine whether a lighter area within the break represents an island of preserved IS/OS or “collapsed” layers, necessitating a review of the corresponding B-scans.
In approximately half of the eyes examined, a dark ring or arc fragments of a ring around the central break were apparent in the “en face” image. In a few cases an additional bright ring was visible. This phenomenon appeared restricted to the IS/OS layer, a corresponding pattern could not be demonstrated in other layers of the retina, and it does not seem to be related to the artifactual ring emanating from specular reflection mainly from the surface of the retina.
46 The origin and significance of this ring are unclear; however, it's similarity in location and shape to the area in which Powner et al. found Müller cell loss or dysfunction within the central macula is remarkable.
47
We found a statistically significant elevation in mesopic sensitivity thresholds within the area of the IS/OS lesion. The difference between mean sensitivities within the break area and the background was just over 8 dB, which is similar to the sensitivity loss reported by other investigators associated with an IS/OS break, despite the differences in underlying disease and methods used.
8,48
Chen et al. found (in 50 predominantly AMD eyes) the coefficients of repeatability for MP1 central macular sensitivity (CMS, inside 10°) to be 2.56 dB, of paracentral macular sensitivity (PMS, 10–20° ring) 2.13 dB and point-wise sensitivity (PWS) 5.56 dB.
49 They recommended the use of CMS and PMS for monitoring macular function, and considered a change of greater than 2.56 and 2.31 dB significant, respectively. Relative to these levels, the differences in our study are significant not just statistically, but also with respect to repeatability. We also found a statistically highly significant positive correlation between IS/OS break area and aggregate sensitivity loss. While the low number of sensitivity test points did not allow a detailed analysis, it was noted that the highest sensitivity loss seemed to be associated with areas of “collapsed layers.”
Traditional imaging modalities (color fundus photography and fluorescein angiography) have limitations. In these, the most prominent features of type 2 MacTel are vascular.
50 Loss of retinal transparency is highly dependent on image quality, and abnormal pigment may accompany otherwise few or mild signs of the disease and also may be absent altogether before neovascularization. These signs also may be challenging to quantify especially in early disease.
New techniques, like SD-OCT (and dual wavelength autofluorescence or blue light reflectance) imaging, have introduced new means for assessing neurodegenerative change in type 2 MacTel. The IS/OS lesion in “en face” SD-OCT images is readily quantifiable and closely associated with function loss such that it may be a prime candidate for a structural outcome measure for following disease progression in the natural history as well as in interventional studies of type 2 MacTel. Further investigations are necessary to determine the progression characteristics of the IS/OS break over time.
Jose-Alain Sahel, Centre Hopitalier National D'Optalmologie des Quinze-Vingts, Paris, France.
Robyn Guymer, Centre for Eye Research, East Melbourne, Australia.
Gisele Soubrane, Clinique Ophtalmolgie de Creteil, Creteil, France.
Alain Gaudric, Hopital Lariboisiere, Paris, France.
Steven Schwartz, Jules Stein Eye Institute, UCLA, Los Angeles, CA.
Ian Constable, Lions Eye Institute, Nedlands, Australia.
Michael Co-oney, Manhattan Eye, Ear, & Throat Hospital, New York, NY.
Catherine Egan, Moorfields Eye Hospital, London, England.
Lawrence Singerman, Retina Associates of Cleveland, Cleveland, OH.
Mark C. Gillies, Save Sight Institute, Sydney, Australia.
Martin Friedlander, Scripps Research Institute, La Jolla, CA.
Daniel Pauleikhoff, St. Franziskus Hospital, Muenster, Germany.
Joseph Moisseiev, The Goldschleger Eye Institute, Tel Hashomer, Israel.
Richard Rosen, The New York Eye and Ear Infirmary, New York, NY.
Robert Murphy, The Retina Group of Washington, Fairfax, VA.
Frank Holz, University of Bonn, Bonn, Germany.
Grant Comer, University of Michigan, Kellogg Eye Center, Ann Arbor, MI.
Barbara Blodi, University of Wisconsin, Madison, WI.
Diana Do, The Wilmer Eye Institute, Baltimore, MD.
Alexander Brucker, Scheie Eye Institute, Philadelphia, PA.
Raja Narayanan, LV Prasad Eye Institute, Hyderabad, India.
Sebastian Wolf, University of Bern, Bern, Switzerland.
Philip Rosenfeld, Bascom Palmer, Miami, FL.
Paul S. Bernstein, Moran Eye Center, University of Utah, UT.
Joan W. Miller, Massachusetts Eye and Ear Infirmary, Harvard Medical School, Boston, MA.